January 20, 1905.] 



SCIENCE, 



85 



Let us consider, first the case of a thermo- 

 electric couple in which neither metal has 

 any Thomson effect, but in which there is 

 a tendency of positive electricity from (1) 

 to (2) lat each junction. The thermo-elec- 

 tric force of such a circuit can be accounted 

 for by assuming that metal 2 attracts posi- 

 tive electricity more and negative elec- 

 tricity less than metal 1, and that both 

 these differential attractions increase or 

 decrease with change of temperature of 

 the electricity. 



At first glance one is likely to think that 

 the differential forces here imagined must 

 increase with rise of temperature, as it may 

 at first seem that the forces at the hot junc- 

 tion must prevail over the opposing forces 

 at the cold junction. But this need not 

 be. The action must be such as to take in 

 heat at the hot junction and to give out 

 heat at the cold junction; but this condi- 

 tion is perfectly consistent with the pre- 

 vailing of the attractive forces at the cold 

 junction. 



For, consider the analogous case of cir- 

 culation of water in a pipe circuit made 

 up of two verticals and two horizontals 

 (see Fig. 5). If heat is applied at the 

 proper part of one vertical and if heat is 

 taken away from the proper part of the 

 other vertical, the water will ascend against 

 the force of gravity at the heated place and 

 descend under the pull of gravity at the 

 cooled place. That is, the attractive force, 

 upon the differential action of which the 

 circulation depends, prevails at the place 

 where heat is taken out from the system. 



Another analogous ease is that of two 

 galvanic cells of precisely the same kind, 

 one cold and the other warm, set to work 

 in opposition to each other. If the cells 

 are such that each M'ould grow warm 

 (aside from the development of resistance 

 heat within its parts) by its own direct 

 action, the cooler cell will prevail, and vice 

 versa. 



So, if the spontaneous action at each 

 junction of our two metals, if each junc- 

 tion could have its own way, would be such 

 as to generate heat at the junction, the 

 cooler junction will prevail when the two 

 are opposed, and vice versa. 



Now we have rather more reason for 

 expecting, in a given untried case, that the 

 free action of attractive forces will generate 

 heat than we have for expecting that it 

 will absorb heat. Consider, for example, 

 the heat freed as the result of molecular 

 attractions in the condensation of a vapor. 

 Accordingly, if we are to account for a 

 thermo-electric current, in such a combina- 

 tion of metals as we have imagined, by 

 attraction of ordinary matter for elec- 

 tricity, this attraction varying with the 

 temperature of the electricity, we are nat- 

 urally led to the opinion that the colder 

 junction prevails. 



The assumption of such an attraction as 

 we have here imagined, with its dependence 

 on the temperature of the electricity and 

 its independence of the temperature of the 

 metal, except as the temperature of the 

 metal determines that of the electricity 

 within it, is much less violent than it at 

 first appears. If there is such a phenom- 

 enon as the expansion of electricity, that is, 

 a diminution of general volume density of 

 electricity, with rise of temperature of the 

 metal containing it, corresponding to the 

 expansion of air or water in the heated 

 part of a convection circuit, this is enough 

 to give just the temperature relation re- 

 quired. For, the lessened volume density 

 of the electricity at the hot junction of the 

 two metals would imply a diminished tend- 

 ency of the electricity to pass over to the 

 more strongly attracting metal at that 

 junction ; but just as there is no tendency 

 of Avater to flow by gravitation along an 

 imequally heated pipe, if this pipe is hori- 

 zontal, so there would be no tendency for 

 electricity to flow along an unequally heat- 



